PREDICTION OF VENTILATOR-ASSOCIATED EVENTS (VAEs)

1. A method of predicting ventilator associated events, comprising: using one or more processors to execute a segmentation module comprising executable code configured to: access time series data of a patient from a ventilator, the time series data comprising one or more features; identify a trigger event in the time series data comprising one or more of a stability period and an oxygen worsening event; and segment the time series data into observation windows based on the trigger event and a set of tuned hyperparameters comprising one or more of observation window duration, observation window overlap, prediction gap, and future event window duration; using the one or more processors to execute a feature extraction module comprising executable code configured to compute a feature-based representation of the segmented time series data, extracting the one or more features for each of the observation windows into a standard format; using the one or more processors to operate a trained classification model configured to evaluate, for each of the observation windows, the one or more features to determine a probability that a ventilator-associated event (VAE) will occur in a future event window associated with a respective one of the observation windows; determining, using the one or more processors, based on output of the trained classification model, that a predicted VAE has a probability of occurring that exceeds a threshold; providing, using an output device operatively coupled to the ventilator, an indication that the predicted VAE is predicted to occur for the patient in the future event window; and proactively providing, based on the indication, a spontaneous breathing trial in advance of the predicted VAE.

2. The method of claim 1, wherein the one or more features include one or more of positive end-expiratory pressure (PEEP) and fraction of inspired oxygen (FiO2).

3. The method of claim 2, wherein the one or more features include one or more of lab results and patient physiological data.

4. The method of claim 1, wherein the indication is one or more of a score, a color, an audible alert, and a recommendation.

5. The method of claim 4, wherein the providing comprises providing data configured to display the indication on a display device associated with one or more of: the ventilator, a patient monitor, a central monitor and a remote device.

6. The method of claim 1, comprising: displaying one or more recommended actions responsive to the indication that the predicted VAE is predicted to occur for the patient in the future event window.

7. The method of claim 1, comprising titrating medicine for the patient based on the predicted VAE.